Generative AI techniques for adapting style of a space

1 . A method for using a generative machine learning (ML) model to generate one or more images of a space in one or more target styles, the method comprising:

using at least one computer hardware processor to perform:

receiving, from a client device,

an image of the space, and

information indicating a target style for the space;

generating, from the information indicating the target style for the space, a textual prompt for use in prompting the generative ML model to generate one or more images of the space in the target style, wherein the textual prompt comprises a plurality of keywords indicating image characteristics to attempt to have in the to-be-generated one or more images of the space in the target style and to attempt to exclude in the to-be-generated one or more images of the space in the target style;

processing the image of the space and the textual prompt by using the generative ML model to obtain at least one generated image of the space in the target style;

detecting at least one furnishing in the at least one generated image;

identifying, in a catalog of furnishing products, one or more furnishing products similar to the at least one furnishing detected in the at least one generated image, the identifying comprising:

obtaining at least one portion of the at least one generated image containing the detected at least one furnishing; and

searching, using the at least one portion and among a set of images of furnishing products in the catalog, for one or more images of furnishing products similar to the detected at least one furnishing; and

sending, to the client device, information about the one or more furnishing products identified in the catalog of furnishing products.

2 . The method of claim 1 , wherein the image of the space is an image of a room.

3 . The method of claim 1 , further comprising:

providing a graphical user interface (GUI) through which a user can provide the image of the space and the information indicating the target style for the space; and

receiving, via the GUI, the image of the space and the information indicating the target style for the space.

4 . The method of claim 1 , wherein the target style is selected from the group of styles consisting of: mid-century modern, coastal, modern farmhouse, Bohemian, industrial, glam and Scandinavian.

5 . The method of claim 1 , wherein the generative ML model comprises a latent diffusion model.

6 . The method of claim 5 , wherein the latent diffusion model is configured to perform text-prompt-guided image-to-image translation.

7 . The method of claim 5 , wherein the latent diffusion model comprises a Stable Diffusion model.

8 . The method of claim 5 ,

wherein the generative ML model further comprises a second trained ML model configured to pre-process the image of the space,

wherein processing the image of the space and the textual prompt using the generative ML model, comprises:

processing the image of the space using the second trained ML model to obtain a pre-processed image of the space; and

processing the pre-processed image of the space and the textual prompt using the latent diffusion model.

9 . The method of claim 8 , wherein the second trained ML model is trained to control the latent diffusion model with task-specific conditions.

10 . The method of claim 8 , wherein the second trained ML model is a trained neural network model configured to detected edges, lines, and/or key points in the image of the space.

11 . The method of claim 8 , wherein the second trained ML model is a ControlNet model.

12 . The method of claim 1 , wherein detecting the at least one furnishing in the at least one generated image comprises:

detecting the at least one furnishing using a trained neural network model trained to perform object detection to obtain at least one portion in the at least one generated image containing the detected at least one furnishing and at least one corresponding label indicating a type for any furnishing so detected.

13 . The method of claim 1 , wherein each particular portion of the at least one portion is a bounding box containing the corresponding furnishing detected in the particular portion.

14 . The method of claim 1 , wherein the searching comprises using a visual search technique to identify, in a first database storing the set of images of furnishing products in the catalog, the one or more images of furnishing products and corresponding one or more identifiers of the furnishing products.

15 . The method of claim 14 , further comprising:

accessing, using the one or more identifiers of the furnishing products and in a second database separate from the first database, the information about the one or more furnishing products.

16 . The method of claim 1 , wherein

sending the information about the one or more furnishing products identified in the catalog of furnishing products comprises:

providing the client device with images of the identified furnishing products and information to facilitate purchase of the identified furnishing products.

17 . A system, comprising:

at least one computer hardware processor;

at least one non-transitory computer readable storage medium storing processor-executable instructions that, when executed by the at least one computer hardware processor, cause the at least one computer hardware processor to perform a method comprising:

receiving, from a client device,

an image of a space, and

information indicating a target style for the space;

generating, from the information indicating the target style for the space, a textual prompt for use in prompting a generative maching learning (ML) model to generate one or more images of the space in the target style, wherein the textual prompt comprises a plurality of keywords indicating image characteristics to attempt to have in the to-be-generated one or more images of the space in the target style and to attempt to exclude in the to-be-generated one or more images of the space in the target style;

processing the image of the space and the textual prompt by using the generative ML model to obtain at least one generated image of the space in the target style;

detecting at least one furnishing in the at least one generated image;

identifying, in a catalog of furnishing products, one or more furnishing products similar to the at least one furnishing detected in the at least one generated image, the identifying comprising:

obtaining at least one portion of the at least one generated image containing the detected at least one furnishing; and

searching, using the at least one portion and among a set of images of furnishing products in the catalog, for one or more images of furnishing products similar to the detected at least one furnishing; and

sending, to the client device, information about the one or more furnishing products identified in the catalog of furnishing products.

18 . At least one non-transitory computer readable storage medium storing processor-executable instructions that, when executed by at least one computer hardware processor, cause the at least one computer hardware processor to perform a method comprising:

receiving, from a client device,

an image of a space, and

information indicating a target style for the space;

generating, from the information indicating the target style for the space, a textual prompt for use in prompting a generative maching learning (ML) model to generate one or more images of the space in the target style, wherein the textual prompt comprises a plurality of keywords indicating image characteristics to attempt to have in the to-be-generated one or more images of the space in the target style and to attempt to exclude in the to-be-generated one or more images of the space in the target style;

processing the image of the space and the textual prompt by using the generative ML model to obtain at least one generated image of the space in the target style;

detecting at least one furnishing in the at least one generated image;

identifying, in a catalog of furnishing products, one or more furnishing products similar to the at least one furnishing detected in the at least one generated image, the identifying comprising:

obtaining at least one portion of the at least one generated image containing the detected at least one furnishing; and

searching, using the at least one portion and among a set of images of furnishing products in the catalog, for one or more images of furnishing products similar to the detected at least one furnishing; and

sending, to the client device, information about the one or more furnishing products identified in the catalog of furnishing products.

19 . A method for using a generative machine learning (ML) model to generate one or more images of a space in one or more target styles, the method comprising:

using at least one computer hardware processor to perform:

receiving, from a client device,

an image of the space, and

information indicating multiple target styles for the space, including a target style for the space;

generating, from the information indicating the multiple target styles for the space, a respective textual prompt for each of the multiple target styles to obtain multiple textual prompts, including a textual prompt for use in prompting the generative ML model to generate one or more images of the space in the target style;

processing the image of the space and the multiple textual prompts by using the generative ML model to obtain generated images of the space in each of the multiple target styles, the processing comprising processing the image of the space and the textual prompt by using the generative ML model to obtain at least one generated image of the space in the target style;

detecting at least one furnishing in the at least one generated image;

identifying, in a catalog of furnishing products, one or more furnishing products similar to the at least one furnishing detected in the at least one generated image, the identifying comprising:

obtaining at least one portion of the at least one generated image containing the detected at least one furnishing; and

searching, using the at least one portion and among a set of images of furnishing products in the catalog, for one or more images of furnishing products similar to the detected at least one furnishing; and

sending, to the client device, information about the one or more furnishing products identified in the catalog of furnishing products.

20 . A system, comprising:

at least one computer hardware processor;

at least one non-transitory computer readable storage medium storing processor-executable instructions that, when executed by the at least one computer hardware processor, cause the at least one computer hardware processor to perform a method comprising:

receiving, from a client device,

an image of a space, and

information indicating multiple target styles for the space, including a target style for the space;

generating, from the information indicating the multiple target styles for the space, a respective textual prompt for each of the multiple target styles to obtain multiple textual prompts, including a textual prompt for use in prompting a generative maching learning (ML) model to generate one or more images of the space in the target style;

processing the image of the space and the multiple textual prompts by using the generative ML model to obtain generated images of the space in each of the multiple target styles, the processing comprising processing the image of the space and the textual prompt by using the generative ML model to obtain at least one generated image of the space in the target style;

detecting at least one furnishing in the at least one generated image;

identifying, in a catalog of furnishing products, one or more furnishing products similar to the at least one furnishing detected in the at least one generated image, the identifying comprising:

obtaining at least one portion of the at least one generated image containing the detected at least one furnishing; and

searching, using the at least one portion and among a set of images of furnishing products in the catalog, for one or more images of furnishing products similar to the detected at least one furnishing; and

sending, to the client device, information about the one or more furnishing products identified in the catalog of furnishing products.

21 . At least one non-transitory computer readable storage medium storing processor-executable instructions that, when executed by at least one computer hardware processor, cause the at least one computer hardware processor to perform a method comprising:

receiving, from a client device,

an image of a space, and

information indicating multiple target styles for the space, including a target style for the space;

generating, from the information indicating the multiple target styles for the space, a respective textual prompt for each of the multiple target styles to obtain multiple textual prompts, including a textual prompt for use in prompting a generative maching learning (ML) model to generate one or more images of the space in the target style;

processing the image of the space and the multiple textual prompts by using the generative ML model to obtain generated images of the space in each of the multiple target styles, the processing comprising processing the image of the space and the textual prompt by using the generative ML model to obtain at least one generated image of the space in the target style;

detecting at least one furnishing in the at least one generated image;

identifying, in a catalog of furnishing products, one or more furnishing products similar to the at least one furnishing detected in the at least one generated image, the identifying comprising:

obtaining at least one portion of the at least one generated image containing the detected at least one furnishing; and

searching, using the at least one portion and among a set of images of furnishing products in the catalog, for one or more images of furnishing products similar to the detected at least one furnishing; and

sending, to the client device, information about the one or more furnishing products identified in the catalog of furnishing products.

22 . A method for using a generative machine learning (ML) model to generate one or more images of a space in one or more target styles, the method comprising:

using at least one computer hardware processor to perform:

receiving, from a client device,

an image of the space, and

information indicating a target style for the space;

generating, from the information indicating the target style for the space, a textual prompt for use in prompting the generative ML model to generate one or more images of the space in the target style, wherein the generative ML model comprises a latent diffusion model and a second trained ML model configured to pre-process the image of the space;

processing the image of the space and the textual prompt by using the generative ML model to obtain at least one generated image of the space in the target style, wherein processing the image of the space and the textual prompt using the generative ML model, comprises:

processing the image of the space using the second trained ML model to obtain a pre-processed image of the space; and

processing the pre-processed image of the space and the textual prompt using the latent diffusion model;

detecting at least one furnishing in the at least one generated image;

identifying, in a catalog of furnishing products, one or more furnishing products similar to the at least one furnishing detected in the at least one generated image, the identifying comprising:

obtaining at least one portion of the at least one generated image containing the detected at least one furnishing; and

searching, using the at least one portion and among a set of images of furnishing products in the catalog, for one or more images of furnishing products similar to the detected at least one furnishing; and

sending, to the client device, information about the one or more furnishing products identified in the catalog of furnishing products.